Compute Deployment Engineer

About this role

Fluidstack is seeking a Compute Deployment Engineer to own the transition of GPU and accelerator compute infrastructure from facility handoff to production readiness. You'll qualify hundreds of racks per site through firmware configuration, burn-in testing, and cluster validation while building automation to scale qualification faster than fleet growth.

What you'll do

• Own compute turn-up from facility availability through production readiness across thousands of racks
• Establish firmware baselines, configure BMC/BIOS, and run burn-in validation on GPU and custom accelerator platforms
• Automate hardware qualification workflows via provisioning stack and Kubernetes to eliminate manual bottlenecks
• Triage hardware failures methodically, isolate faults to component level, and drive RMA/vendor escalation
• Partner with network, ICT, data center ops, and hardware teams during turn-up windows and incident response
• Travel to data halls for weekly on-site turns during new facility bring-up phases

What they're looking for

• Linux administration and out-of-band management (BMC, IPMI, Redfish)
• Python or Go for infrastructure automation
• Large-scale server or GPU fleet deployment and production bringup
• Hands-on data center experience: racking, cabling, component troubleshooting
• Hardware failure diagnosis across firmware, software, and physical layers
• Kubernetes-based bare-metal provisioning (bonus)
• GPU/accelerator platform bringup experience (bonus)
• DCIM and inventory systems (bonus)

Benefits

• Work on civilization-scale AI infrastructure
• Extreme ownership with full autonomy over end-to-end projects
• High-velocity environment pushing technical frontiers
• Competitive compensation with pay equity commitment
• Mix of remote and on-site work with structured travel windows
• Collaborative team spanning hardware and software domains

Likely interview questions

• Describe the largest server or GPU fleet you've brought to production—how many nodes, what was the critical path, and what bottlenecks did you hit?
• Walk us through a time you diagnosed a hardware failure across firmware, software, and physical layers. How did you isolate the root cause?